Skip to content

Tags: pengchujin/jzsub

Tags

v1.8.1

Toggle v1.8.1's commit message
[verified] Improve portrait subtitle safe area

v1.8.0

Toggle v1.8.0's commit message
Make the translation target language configurable

- --target-lang (default zh-CN) flows from the fetcher through the
  manifest into translation batches, rendering, and validation
- Source-track exclusion follows the target: a ja target skips ja
  tracks and accepts Chinese sources
- Translation contract v4 declares target_language and a generic
  "translation" output field; legacy zh_cn manifests still validate
- Chinese punctuation house style applies only to Chinese targets;
  the translated SRT is named after the target language tag
- Document --target-lang in SKILL.md, the contract, and README

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

v1.7.0

Toggle v1.7.0's commit message
Add --deliver targets for video-only and subtitle-only jobs

- --deliver full|video|subs|bilingual-subs selects the delivery
  target; the manifest records it and verify_delivery.py gates
  completion against the declared target
- video keeps downloading source subtitle files but skips the
  translation pipeline; subs/bilingual-subs skip video streams and
  fail clearly when no suitable subtitle exists
- subtitle-only jobs take caption layout size from probed metadata
  instead of local media
- document the targets in SKILL.md intent routing and README

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

v1.6.1

Toggle v1.6.1's commit message
Draw one background panel per caption with BorderStyle=4

Per-line BorderStyle=3 boxes overlap between adjacent lines and
double-paint the translucent color into dark bands. libass
BorderStyle=4 draws a single panel behind the whole stacked event;
the burn path already requires libass, so the extension is safe.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

v1.6.0

Toggle v1.6.0's commit message
Widen Chinese lines and drop sound-annotation cues

- Give Chinese the same pixel budget as the source language (62
  columns / 31 CJK characters at 1920) so short sentences no longer
  wrap needlessly; portrait still narrows with available width
- Exclude cues that are only sound descriptions ([Music], [Applause],
  【音乐】, (拍手), music notes) from the ledger, translation, and
  render; mixed dialogue cues stay untouched and Japanese 「」quotes
  are never treated as annotations
- Degrade annotation-only subtitle tracks to a video-only delivery
  (dialogue:false in the manifest) instead of failing the pipeline
- Bump subtitle pipeline version to 1.1 so stale manifests fail fast

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

v1.5.0

Toggle v1.5.0's commit message
Unify subtitle translation and fix caption anchors

v1.4.0

Toggle v1.4.0's commit message
Optimize compact translation workflow

v1.3.0

Toggle v1.3.0's commit message
JZSub v1.3.0

v1.2.0

Toggle v1.2.0's commit message
JZSub v1.2.0

v1.1.1

Toggle v1.1.1's commit message
JZSub v1.1.1